Built to Boom: Why the AI Bubble Isn’t Popping
The dot-com crash laid the groundwork—and the cash piles—that are keeping the AI economy afloat
The last few weeks have been very unsettling as it pertains to understanding the idea of an AI bubble. The global financial world has been turned upside down by the current administration’s trade policies. This sort of event normally treats all sectors equally with capital fleeing from risk into the safest harbors. And that’s what happened, for the most part. The Dow Jones and S&P 500 both took a major hit as did the bond market. Many expect this to be just the beginning. There is significant concern about inflation returning and longer-term worry that the U.S. may have left itself out in the cold in terms of world trade. Yet amidst all this chaos, the AI category just kept sailing. This is important if for no other reason than it begs the question of, why? Why does a technology category that many have called a speculative bubble get a pass when the rest of the world gets rocked?
Not all AI economy companies were immune. NVIDIA announced it would take a $5.5 Billion hit due to policies. Yet, at the same time, OpenAI’s former CTO Mira Murati announced she was opening Thinking Machines Lab with an opening valuation of $10 Billion (despite no IP or products), Andreesen Horowitz announced a $20 Billion mega round and OpenAI co-founder Ilya Sustkever announced the formation of Safe Superintelligence Inc. (SSI) with an opening valuation of $32 Billion.
Market math doesn’t square with conventional thinking or historical norms. History teaches us how business and technology exists in boom and bust cycles. Analysts and scholars are increasingly wary of Black Swan Events that might turn a boom into a bust. The current trade war is just such an event. Yet the AI category not only persists, it prospers. It’s as if it’s completely immune. The mass proliferation of start ups and new ventures with eye-popping multiples and valuations continues, unabated. Many believed rising interest rates would cause a market slowdown. The era of cheap money is over. Yet 2024 was the largest AI funding year to date. All these economic headwinds would traditionally cause market pullback.
There can only be one of two possible explanations for why this is:
AI is truly so promising and world changing that it defies logic and reasoning.
This investment cycle is structurally different - investors and entrepreneurs have actually learned from past mistakes and have approached business differently this time.
While AI pundits and founders would like you to think it’s #1, the facts tell you it’s actually more likely #2.
The bubble metaphor
Up until recently the investment bubble metaphor was perfect. It aptly described the market dynamic. The market itself would continue to expand until such time as either the surface tension became too much or some external force caused it to tear. You would watch it and watch it but you could never know exactly what it might pop. There are those moments when you think it will but then it stabilizes. Then, pop. The bubble is no more.
And up until now, the AI bubble looked just like the dot com bubble to most everyone - the difference being how it weathered the past few weeks. This begs the question of how the AI category is different. When you look at the dot com and AI bubbles side by side you see three key differences that help explain. They are as follows
Investors have become unicorn ranchers: Investors are obsessed with unicorns. This is a shift from traditional investment strategy. Unicorns are companies that quickly reach at least a $1Billion valuation. CB Insights tracks venture investment. In its State of AI Q4 '24 report it shared how 64 percent of AI unicorns were in either the validating or deploying phases meaning they were in the early stage. This is as opposed to 81 percent of non-AI startups being later phase companies (scaling or established). Investors are pumping more and more cash into a pool of early stage companies allowing them to hold off on pushing to IPO. Which brings us to point number two.
IPOs are the exception: During the dot com era, an early IPO was the goal for most investment plans. There were over 350 tech IPOs in 1999 alone. The AI boom is several years in yet we have seen very few. Investors have a longer-term business plan this time around which has heretofore shielded AI companies from the kind of pressure and transparency that can bring a company down. The combination of rush to unicorn and refrain from the markets allows investors to maintain stratospheric valuations based on their own assessment. But that doesn’t explain the seemingly bottomless piles of cash fueling these valuations.
Dot Com survivors + 2024 Front Loading: While most dot com darlings went bust, some not only survived but grew into near monopolies with huge cash cushions. Amazon and Google are two such companies. Microsoft and Apple may not have been traditional dot com companies but they could have been toppled all the same. Each of these companies sit on hundreds of billions of dollars in cash and are putting that cash to work to build out AI infrastructure in hopes of owning a sizable piece of the future. This essentially creates a secondary line of equity credit for the AI category that hasn’t existed for other cycles. Corporate investment is different from equity market investment as it can be managed and written down if it doesn’t work out, buried as a line item in the 10-k. You also have to take into account how Q4 '24 was the single largest AI investment period ever, likely front loading investment strategies in anticipation of a challenging 2025. Q4’24 saw almost $44 Billion in AI VC investment according to CB Insights. This is compared to ~$56 Billion in the first three quarters. While it’s impossible to say definitively, it’s easy to assume funds wanted to put their capital raises to work quickly before things got hairy.
It’s impossible to say what, if anything, will happen to the AI category. At this point it’s as likely to go up in flames as it is to become the future. But if it survives and prospers experts will likely point out how so much of it was built on the foundational work created during the dot com era.
The success of the AI era didn’t materialize out of thin air—it was built on the smoldering remains of the dot-com bust. The fiber optic networks, server farms, and cloud systems that now power large-scale model training were originally born from the excess and ambition of that earlier time. What was once overbuilt and underutilized became the bedrock of modern digital infrastructure. The companies that survived—Amazon, Google, Microsoft—used the decades since to consolidate power, scale operations, and hoard cash. Now they’re redeploying those reserves at historic levels, making AI less a speculative gold rush and more a cold, calculated land grab by the tech titans. Their investment isn’t passive—it’s strategic, long-term, and often invisible to the public markets. This isn’t about quarterly earnings. It’s about owning the future.
That’s the real story behind the apparent irrationality of AI’s market resilience. While many point to frothy valuations and assume we're reliving the late ’90s, the differences matter. This time, companies are delaying IPOs, staying private longer, and leveraging the runway gifted to them by both venture capital and corporate giants. It’s not that we’ve avoided a bubble—it’s that the underlying support structure is stronger, thicker-skinned, and better capitalized than the dot-com era ever was. The AI boom may still burst. But if it doesn’t, it’ll be because the last bubble built the infrastructure this one now thrives on.